[aur-general] Proposal: Mass AUR Cleanup

[Peter Lewis]

On Sunday 03 October 2010 at 21:51 Isaac Dupree wrote:
> >> I've just created a new proposal concerning the orphaning of all
> >> packages marked 'out of date' which have not been updated (or
> >> submitted) since before January 1st, 2009. For details, see the actual
> >> proposal text.
> > 
> > Jakob, would it be possible for you to parse the e-mail address of the
> > current maintainer, and for each package send an e-mail? The template
> > could look like this:
> > 
> > """
> > Subject: [AUR][$pkgname] Orphan candidate
> > 
> > Hi
> > 
> > Your package '$pkgname' [1] is a candidate to be orphaned, as per the
> > out-of-date and inactivity list generated recently [2]. If you do not
> > update '$pkgname' by the 10th of October 2010, it will be
> > automatically orphaned.
> > ...
> 
> maybe just something like, "If you wish to continue maintaining it and
> are able to do so, please re-adopt and update it or state why this
> cannot be done?"
> 
> ...Orphaning generally isn't a very harmful act.
> 
> And/or we could go by the one-by-one orphaning principle of giving
> maintainers a couple weeks to respond to email.  It's likely many of
> these maintainers won't respond because they've been inactive on AUR for
> a year and a half; and if they do respond, it's evidence that they're
> active enough to warrant a not-so-automatic process perhaps?
> 
> (So the critical date of orphaning then should/would probably be later
> than 10th October.)

Yeah. Though I really do appreciate what Jakob is doing, I like the idea of 
using email to notify maintainers first. It seems that if someone's gone to 
the trouble of uploading something to our repo at some point, then IMO they at 
least deserve an email when it's being orphaned or deleted. But, obviously 
this is a reasonable amount of work without some sort of automating process.

So regardless of whether or not we go ahead with the proposal as is now, I 
think we should try to sort out some kind of regular (semi-)automated process 
for this in the future.

Also, once timestamps for package out-of-date flagging are up and running, 
this will be much easier to automate, I would imagine. We could even have 
something that reminds/prompts/warns the maintainer n weeks after the initial 
flagging, and then orphans it m weeks later if it's not unflagged.

Nice work Jakob, BTW :-)

Pete.